How to display data badly

Link: Wainer_1984 Please write a review of “How to display data badly” article by Wainer. Your review should be between 300-500 words, including important guidelines of ways showing your data efficiently. Please post your answers as a reply. Due date March 24,  2015.

 

11 thoughts on “How to display data badly

  1. Wainer aimed to examine different types of methods that display data in a poor and uninformative way. In this article he divulges 12 rules to making bad data displays. First, he explains that in order for data graphics to be good, it must be able to display data accurately and clearly. Flawed data graphics either have error in showing data, showing data accurately, or in showing data clearly. He argues that the first rule in showing data poorly is to minimize the data density by showing as few data as possible. When poor data displays contain little information, they look empty and one will try to fill in the space with “chartjunk.” This leads to the next principle of bad data displays, hide what data you do show by minimizing the data-ink ratio. The data-ink ratio measures the amount of ink used in the data to the total amount of ink in the graph. Bad graphs have data-ink ratios close to 0. One is capable of hiding data in the grid or scale of a graph. Thus, data can be displayed poorly by either not including the data or by hiding it.
    Another way to display data poorly is to ignore the visual metaphor that represents the data and to create displays where only order of the numbers matter and their magnitude is not accounted for. Furthermore, often people create poor display data by graphing data out of context. These ways specifically affect showing data accurately. He then transitions into explaining ways people make mistakes in showing data clearly. Specifically, changing scales in mid-axis results in poorly displayed data and even misleads the reader. In addition to this, creating graphs that emphasize the trivial data and not the important information makes it unclear what the point of the research was and may distort the actual important information.
    Furthermore, Wainer stresses that it is important to start from the same base when making comparisons. He explains that poor graphs start from different bases. One should also never alphabetically order graphs and tables. Instead figures should be ordered based on some aspect of the data. In addition to the order of figures, how data displays are labeled are vital in communicating the right idea to the reader. Poor figures often have labels that aren’t descriptive enough. Another important point that Wainer stresses it that too many decimals can cause the figure to be difficult to understand. Finally, Wainer suggests that if a data display has previously been done well, then to think of another way to do it. Although the article was light hearted, Wainer stresses some important concepts to keep in mind when creating figures to represent data.

  2. In order to be able to present good data, you must know how to recognize bad data. Wainer (2001) defines the aim of good data as graphs that display data accurately and clearly. He further breaks this definition into three parts: (a) showing data, (b) showing data accurately, and (c) showing data clearly.
    Showing the data includes showing enough of it. If holding clarity and accuracy constant, good data graphics contain enough data points to efficiently convey the information. In addition, the data being shown in the graph needs to be an appropriate scale with no “chartjunk” making around it to confuse a reader.
    Showing data accurately means to show a graphic display that has both magnitudes and an order and that is represented by an appropriate visual metaphor (Wainer, 2001). This includes not changing the scale of the numbers during the graphic and using the right starting point- not after a big drop in data or after you zoomed in on a single meander, for example. In addition to this, do not use perceived length as a way to represent data.
    It is also important to show the data clearly. For example, do not change the scale mid-axis. Make sure to account for the trivial information gathered from data, but focus on the larger and more question focused data. Another point to ensure that the data presented is good is to never “jiggle the baseline.” Meaning, make sure that each of the data points has its own baseline and is not being added to the one below it. This is important for clearly showing the trends of the data.
    Alphabetizing the data (countries for example) is not a good way to represent the data. Order it in the most clarifying way to present the data. Also, do not add unnecessary decimal places, this just confuses the visual representation. These two are the last points to take into consideration when displaying data.
    In summary, it is vital to present the data you have clearly. By showing the data, showing it accurately, and showing it clearly, you can then effectively show the reader what you are trying to convey.

  3. Performing studies, collecting data, and displaying the data for analysis is not something that has recently emerged in society. Similarly to the diverse ways in which data can be shown (graphs, charts, and plots, etc.) there is also many ways in which data can be showed incorrectly. In order to ensure that the data presented gets across to the reader it must always be presented accurately and clearly. This may seem rather simple but there are a number of things to keep in mind when trying to effectively exhibit information. One good rule of thumb is to first make sure that you have enough information to illustrate. While the information can be completely accurate, empty illustrations from lack of material can prove to be a hinderance to overall understanding. Another popular error is to treat data like a piece of art and use vibrant colors and images. While this may make the information more pleasing to the eye, it only adds distractions for the viewer. Again, this can result in confusion and complete misunderstanding. However, information can be both easily understood and aesthetically pleasing by using diagrams that consist of clear-cut lines, values, and descriptions that do not clutter the data. Uniformity is yet another important characteristic of displayed data. For example, it is not advised to alter the scale in the middle of a graph so that slightly varying values can be includes into the same diagram. The reason for this is that very large and significant differences can be undermined when the reader compares them to values that are plotted on a different scale. Continuing on, the type of diagram in which information is displayed determines whether or not your information looks meaningful. For instance, displaying numbers that are relatively close in a bar graph may not show any difference amongst different samples. However, if a line graph or a stem-and-leaf diagram are utilized it may be much clearer that to the reader that there are radical differences amongst variables in the study. In summary, information can be shown through many different techniques. However, no matter how different the method is, there are certain approaches that should always be considered so that the aims of the study are clear to the reader.

  4. The article “How to Display Data Badly”, by Howard Wainer, is a phenomenal tool for students that are learning how to organize and display data in scholarly research. Since scholarly literature is typically quite dense, the goal of showing that your research is worth paying attention falls to the visual descriptions of data. The article is broken into three main sections that represent three arenas of data display that could be messed up, showing data, showing data accurately, and showing data clearly. Within these three subsections there are twelve rules that create bad data displays. When discussing the first section, the following rules are mentioned: Show as few data as possible and Hide what data you do show. The previous two rules are stating that while having too much data is over stimulatory the correct amount is not almost nothing. Similarly, if data is important make sure it is addressed; data that is not important or significant should not be tackled more. In the second section about showing data accurately these rules are discussed: Ignore the Visual Metaphor altogether, only order matters and Graph data out of context. These rules are stating that for some data, reorganizing is not necessary as it is already displayed properly or that magnitude is more important than the order of data. Also, graphing data but putting it out of context can change the perceptions of the data significantly and cause the audience to reach different conclusions. For the final section the remaining rules are involved: Change scales in mid-axis, Emphasize the trivial, Jiggle the baseline, Austria first, Label illegibly, incompletely, incorrectly, and ambiguously, more is murkier, and if it has been done before do it another way. For these final rules, the main focus is to present the data in ways that are easy to follow draw conclusions from. If the titles of axes or measurements are labeled incorrectly, then it will be difficult for people to draw conclusions and understand why it is important. Overall this was a fantastic article and it got its point across clearly and accurately; however, since they were rules for what to do if you wanted to display things poorly it was hard to follow.

  5. In How to Display Data Badly, Howard Wainer defines three guidelines for data representation: show your data, show your data accurately, and show your data clearly. Under each guideline, Wainer discusses general rules (12 in total) for how to badly display and therefore misrepresent data within these guidelines. Under the guideline showing your data he discusses two rules. He mentions the data density index (DDI), which is defined as the number of numbers, plotted per square inch. Rule 1 is show as few data as possible and minimize data density. He warns readers to beware of chart junk, which use a lot of ink in the data but do not contribute to the actual findings being reported. Rule 2 is to hide what data you do show. When properly displaying data it is important to ensure that sufficient data is represented and that your data is visible and not hidden.

    Under the guideline showing your data accurately three rules are mentioned for poor data representation. Rule 3 is to ignore the visual metaphor. To properly display data one must not shuffle the relationship of natural order within the data. Rule 4 is only order matters. Here Wainer warns not to only focus on order. Data distortion by inaccurate length displays should be avoided. Rule 5 is to graph data out of context. It is important to make sure that data is displayed at appropriate intervals.

    The last set of rules fall under the guideline, show your data clearly. Rule 6 is to change the scales in mid axis. To properly display data one must consistently scale data. Rules 7 is emphasizing the trivial and ignore the important. It is crucial to focus on and not distract from the important information in your data. Rule 8 is jiggle baseline. Bad data representation starts from different bases. Make sure to compare all data from the same baseline. Rule 9 is Austria first. Some bad data displays show data alphabetized. It is more important to order data based off of some aspect within the data rather than by alphabetical order. Rule 10 is label illegibly, incompletely, incorrectly, and ambiguously. It is crucial to properly label data so as to ensure that readers understand what is being represented. Rule 11 is more is murkier. Reporting too many decimal places or using too much color will distract readers from what the actual point of the data is. Rule 12 is if it has been done well in the past, think of another way to do it. On the contrary if data has been successfully represented in certain way, do not try and reinvent the wheel. Follow the representation as a template.

    Overall, this article listed many important points in which data can be misrepresented. It was helpful for gaining not only a better understanding of how to better represent data but also how to evaluate data already reported.

  6. “How to Display Data Badly,” an entertaining satire that details the common shortcomings of data display, raises some relevant and interesting points pertaining to effective (or ineffective) data displays. The distinctions between “showing data, showing data clearly, and showing data accurately” are further examined; in each category, various guidelines for displaying data badly are established and examples of each are provided. Data can be arranged such that the amount of information actually displayed is kept to a minimum and is not representative of entire sample from which data was collected. Therefore, any conclusions drawn from the data are not entirely reliable. Even if all data is included, it can be displayed in a manner that does not allow for correct interpretations of trends or patterns. Adding superfluous graphics or unrelated data can distract from the true meaning and applicability of the data. Bending and manipulating data to confine the meaning of the data to predetermined conclusions or interpretations also ruins the integrity of the true data. Although many of the pointers for displaying data poorly are easy to define and identify, such as Rule 6 (Change Scales in Mid-Axis), Rule 1 (Show As Few Data As Possible), or Rule 9 (Austria First!), other rules such as Rule 5 (Graph Data Out of Context) and Rule 11 (More is Murkier) are much less obvious and could even be unconsciously applied during the creation of data displays. Moreover, many of the rules provided raise ethical questions regarding the manipulation of data displays to alter how the significance of the data is perceived. To create bad data displays, some researchers probably make conscious choices that violate the very standards by which all scientists must abide: to discover the truth of scientific processes and report it as such. Reviewing this article gave me the skills to carefully examine data displays and analyze for any tell-tale signs relating to displaying data badly.

  7. Wainer’s article, focusing on how to display data badly, outlines twelve effective ways to misrepresent data on graphs. It is extremely important to effectively and efficiently display data, because it is heavily examined when a paper or article is read. Data has a way of showing what you are talking about, making it, potentially, an effective tool for your readers. Wainer breaks up the twelve ways into three sections: showing data, showing data accurately, and showing data clearly.
    The first section, showing data, had two main points: minimize the data density and hide the data you do show. By limiting or excluding certain data points from your graph, you are able to manipulate what your graph “says” about your data. Overall, your data density should be a bigger number because it ultimately reflects that you are displaying more data. By hiding the data you do have, with “chart junk” and other fillers, the main point of the data is lost.
    The second section, showing data accurately, had four main points: ignore the visual metaphor altogether, only order matters, and graph data out of context. By ignoring the visual metaphor altogether, the overall “image” or understanding of the graph is misconstrued. Only order matters reflects when a trick is used, using length as a visual metaphor. The importance becomes placed on numbers and not the magnitude of data. When data points are chosen, manipulated, or taken from particular parts of the data, data tends to be graphed out of context, thus the overall meaning of the graph is different.
    The last section, showing data clearly, had six main points: change scales in mid-axis, emphasize the trivial, jiggle the baseline, alphabetical order, labeling to confuse, more is murkier, and if it has been done well in the past, think of another way to do it. Changing scales in mid-axis distorts the overall data and manipulates what the author of the graph wants the data to say. Emphasizing the trivial draws the reader into the less important parts of the graph, leaving out what really is important. To jiggle the baseline means to make your comparison graphs confusing by starting form different bases. Alphabetical order, or any particular way data is structure, upsets the order of presentation, manipulating what the data will show on a graph. Labeling to confuse the audience can be done through making it illegible, incomplete, incorrect, and ambiguous; therefore, the readers are unaware of what the data may really be showing. More is murkier is the idea that when more decimal places and more dimensions that are used, the more confusing the graphs become to interpret. Lastly, if a graph has been done a certain way in the past, and well, that when you do it another way it sets your data apart.
    While this article highlighted all the things you can do wrong with your data, I found it extremely helpful on what I should be looking for when I put my data together or when I look at others’ data. The writing overall was a bit confusing, but I understood the gist of what the author was saying and can easily use this as a guide of what not to do.

  8. In Wainer’s analysis entitled “How to Display Data Badly”, he emphasized 12 separate ways in which data can be presented in a misleading and inefficient way in order to encourage better practices in data display. Wainer first discusses the consequences of showing as few data points as possible. Not including data points that are essential to analysis can lead to incorrect conclusions. He then discusses the importance of using good technique when one plots data, by using appropriate grid lines and scale. His next two tips are related to visual metaphors. He points out how important shading, size, and time scales can be, as well as the order of the visual metaphors. Wainer warns against graphing data out of context. Important data can often be left out if the presenter chooses to focus on a specific interval that does not include these data points. He also cautions against changing scales in mid-axis as this can have a profound impact on the way data is interpreted by making large changes in data look less significant and vice versa. Another important point that Wainer discusses is that one should not emphasize the trivial aspects of the data while ignoring the most important findings. There are several ways to do this with the techniques he previously discussed, and should be avoided. He refers to jiggling the baseline any time one makes comparison to the control or base unclear. Another tip he gives is to try to label graphs and tables by trend or some other related factor as opposed to simply listing alphabetically or in another way that also confuses comparison. He gives a few cautions in terms of labeling, warning his readers to always make sure they are labeling legibly, completely, correctly, and unambiguously. Wainer also takes the time to discuss the potential negatives involved in the inclusion of extraneous detail into the data, such as an overwhelming number of decimal points or a huge amount of variables in one graph. Lastly, he encourages his readers to learn from example. If a graph looks particularly good and represent data exceptionally, then don’t diverge from this method! Overall, Wainer’s points are all very valid in creating an effective guide for how one can accomplish presenting data well.

  9. In this article Wainer discusses twelve ways in which data can be displayed badly. By making note of the ways in which data should not be constructed and displayed, it can be a fundamental guide to the ways in which statisticians as well as other scientists should go about conveying data to their intended audiences. The three aims of good data graphics are to display data, show data accurately, and show data clearly. Therein these are the three routes by which data can be misconstrued. One of the methods by which data is displayed poorly is to show as little data as possible by minimizing the data density. This is not to say that the higher the data density, the more efficient the transmission of information is. The information also needs to be accurate and have clarity, but an asset of graphical displays is that a large amount of information can be conveyed in such a small space. The second rule of bad data is that the data that is shown is hidden. As well as when visual metaphor is ignored and not consistent throughout the entirely of the graph this can be misleading to the viewer. Graphing data out of context also makes it hard for the viewer to correctly interpret the meaning of data and can often cause the viewer to be misled. These methods are ways in which data is not displayed accurately which causes an inability to interpret and even understand the information present. However, there are other ways that accurately displayed data can be subtly altered to hinder the portrayal of the most meaningful or interesting information. One of these methods is to change the axis scales mid-axis, which creates a skewed scale by which large changes can look small and exponential ones can look linear. Also, when trivial information is emphasized more than the important point of the graph, this is an erroneous method that should be avoided when constructed a good graph. Another error that should be avoided is starting from different baselines with each quantities which is the only way in which a fair comparison of quantities can be made. Data should never be ordered alphabetically, as this can misconstrue the data that could have been clearly ordered based upon an intrinsic quality of it. Wainer further goes on to say that data always needs to be labeled legibly, completely, correctly, and unambiguously. Less is more in several aspects of data, especially in regards to using an ample amount of decimal places as well as not creating too many dimensions to the data, which only serve to muddle the viewer’s perception. As an individual trying to create a graphical interpretation one should try to think of other various methods of doing things that have already been done well. By following these policies, one can learn how to accurately and clearly portray data to an audience and not make the same mistakes that Wainer satirically champions in Minard’s graph of the French army.

  10. Wainer’s “ How to Display Data Badly” discusses fundamental wrong and misuse of data displaying techniques. Ironically, while the article was published in 1984, I found the article to still applies today since many of the problems discussed in the articles can still be found in the media, school, or even certain published articles. In the article, Wainer pointed out three essential elements that would make data display good. These three essential elements includes 1) the display must actually shows the data. 2) The data must be display accurately. 3) The data must be displayed clearly.The first element is very simplistic, thus, the second and third elements: displaying data accurately and clearly would be discussed in the context of the twelve rules according to Wainer.
    According to Wainer, in order to display the data accurately, first of all, the data display must not hide the data that needed to be displayed. This means that the data should not be blown of scale or leave out parts that would result in altering of how data would come out. For instance, for a line graph, zero must always be included. Secondly, similar to the first rule, the visual metaphor of the data must be accurate and proportionate; otherwise, the trend shown would not be accurate. Figure 6 in the article is shown as a bad example of how the visual affected the accuracy of the scale. Thirdly, the data should not be taken out of context. This means that the data should not be cut or leave out important label that would actually show what the data really is; Figure 11 in the article is an example of how data could be distorted by changing the context of the data. Overall, in order to display data accurately, scale, baseline, and label must be consistent and corrected.
    Besides accuracy, Wainer also mentioned the importance of clarity in which the data are displayed. Personally, I think this is really the main problem that many students have today; many students utilize “visual junk” such as 3D bar graph or very similar color options in bar or line graphs. In addition to “visual junk”, Wainer mentioned how visually could trick the eyes, thus cause bias in data interpretation by reader. Figure 23 is an example of this. Other than these two qualities, organization is also very importance. For instance, the data should be arranged in orderly manner (e.g. from big to small, increasing in quantity, etc.). As for display of numeric value, Wainer mentioned that the decimal places should be kept just sufficient enough to show the accurate differences between variables, since too many could make trends harder to be observed.
    Overall, I feel like by following these rules students could make sure that could display data well, while at the same time, be more intuitive at picking out dishonesty and poor data presentation by others

    • I noticed that the time indicating when these comments are posted are not correct, so I just want to state that this comment is posted on March 24th at 11.17 PM

Leave a Reply